[jira] [Comment Edited] (YARN-8549) Adding a NoOp timeline writer and reader plugin classes for ATSv2
[
https://issues.apache.org/jira/browse/YARN-8549?page=com.atlassian.jira.plugin.system.issuetabpanels:comment-tabpanel=16758740#comment-16758740
]
Vrushali C edited comment on YARN-8549 at 2/1/19 10:43 PM:
---
Hi Prabha
Can you update the patch for branch-2 as well to make it in sync with the trunk
patch? I tried applying the latest trunk v5 patch to branch-2 but it does not
compile.
{code}
[INFO] BUILD FAILURE
[INFO]
[INFO] Total time: 02:35 min
[INFO] Finished at: 2019-02-01T14:34:11-08:00
[INFO]
[ERROR] Failed to execute goal
org.apache.maven.plugins:maven-compiler-plugin:3.1:compile (default-compile) on
project hadoop-yarn-server-timelineservice: Compilation failure: Compilation
failure:
[ERROR]
/Users/vchannapattan/Vrushali/code/Q22017_hadoop/hadoop_commit/hadoop/hadoop-yarn-project/hadoop-yarn/hadoop-yarn-server/hadoop-yarn-server-timelineservice/src/main/java/org/apache/hadoop/yarn/server/timelineservice/storage/NoOpTimelineWriterImpl.java:[22,58]
cannot find symbol
[ERROR] symbol: class TimelineDomain
[ERROR] location: package org.apache.hadoop.yarn.api.records.timelineservice
[ERROR]
/Users/vchannapattan/Vrushali/code/Q22017_hadoop/hadoop_commit/hadoop/hadoop-yarn-project/hadoop-yarn/hadoop-yarn-server/hadoop-yarn-server-timelineservice/src/main/java/org/apache/hadoop/yarn/server/timelineservice/storage/NoOpTimelineWriterImpl.java:[63,38]
cannot find symbol
[ERROR] symbol: class TimelineDomain
[ERROR] location: class
org.apache.hadoop.yarn.server.timelineservice.storage.NoOpTimelineWriterImpl
[ERROR]
/Users/vchannapattan/Vrushali/code/Q22017_hadoop/hadoop_commit/hadoop/hadoop-yarn-project/hadoop-yarn/hadoop-yarn-server/hadoop-yarn-server-timelineservice/src/main/java/org/apache/hadoop/yarn/server/timelineservice/storage/NoOpTimelineWriterImpl.java:[61,3]
method does not override or implement a method from a supertype
[ERROR] -> [Help 1]
[ERROR]
{code}
thanks
Vrushali
was (Author: vrushalic):
Hi Prabha
Can you update the patch for branch-2 as well to make it in sync with the trunk
patch?
thanks
Vrushali
> Adding a NoOp timeline writer and reader plugin classes for ATSv2
> -
>
> Key: YARN-8549
> URL: https://issues.apache.org/jira/browse/YARN-8549
> Project: Hadoop YARN
> Issue Type: Sub-task
> Components: ATSv2, timelineclient, timelineserver
>Reporter: Prabha Manepalli
>Assignee: Prabha Manepalli
>Priority: Minor
> Attachments: YARN-8549-branch-2.03.patch,
> YARN-8549-branch-2.04.patch, YARN-8549.v1.patch, YARN-8549.v2.patch,
> YARN-8549.v4.patch, YARN-8549.v5.patch
>
>
> Stub implementation for TimeLineReader and TimeLineWriter classes.
> These are useful for functional testing of writer and reader path for ATSv2
--
This message was sent by Atlassian JIRA
(v7.6.3#76005)
-
To unsubscribe, e-mail: yarn-issues-unsubscr...@hadoop.apache.org
For additional commands, e-mail: yarn-issues-h...@hadoop.apache.org
[jira] [Comment Edited] (YARN-8549) Adding a NoOp timeline writer and reader plugin classes for ATSv2
[ https://issues.apache.org/jira/browse/YARN-8549?page=com.atlassian.jira.plugin.system.issuetabpanels:comment-tabpanel=16632209#comment-16632209 ] Vrushali C edited comment on YARN-8549 at 9/28/18 6:21 PM: --- [~suma.shivaprasad] Any particular reason you want would like null to be returned instead of a new TimelineWriterResponse? I think if we return null, it may result in a NPE instead of silently dropping the data on the floor. [~prabham] One thing that I am wondering if it might be useful is that perhaps we should add a debug log message saying that NoOP writer has enabled, hence dropping the data? It may be hard to figure out the data loss if a particular machine has this enabled but rest of the machines have the regular writer enabled. In that case, partial data will be missing and the logs wont tell us that no-op writer has been enabled. So perhaps in debug mode at least, we should log for writes to timeline service that no op writer is enabled. Also for reader, else we won't know why no data is being returned if one reader instance has this enabled and other reader instances don't. was (Author: vrushalic): [~suma.shivaprasad] Any particular reason you want would like null to be returned instead of a new TimelineWriterResponse? I think if we return null, it may result in a NPE instead of silently dropping the data on the floor. One thing that I am wondering if it might be useful is that perhaps we should add a debug log message saying that NoOP writer has enabled, hence dropping the data? It may be hard to figure out the data loss if a particular machine has this enabled but rest of the machines have the regular writer enabled. In that case, partial data will be missing and the logs wont tell us that no-op writer has been enabled. So perhaps in debug mode at least, we should log for writes to timeline service that no op writer is enabled. Also for reader, else we won't know why no data is being returned if one reader instance has this enabled and other reader instances don't. > Adding a NoOp timeline writer and reader plugin classes for ATSv2 > - > > Key: YARN-8549 > URL: https://issues.apache.org/jira/browse/YARN-8549 > Project: Hadoop YARN > Issue Type: Sub-task > Components: ATSv2, timelineclient, timelineserver >Reporter: Prabha Manepalli >Assignee: Prabha Manepalli >Priority: Minor > Attachments: YARN-8549-branch-2.03.patch, > YARN-8549-branch-2.04.patch, YARN-8549.v1.patch, YARN-8549.v2.patch, > YARN-8549.v4.patch > > > Stub implementation for TimeLineReader and TimeLineWriter classes. > These are useful for functional testing of writer and reader path for ATSv2 -- This message was sent by Atlassian JIRA (v7.6.3#76005) - To unsubscribe, e-mail: yarn-issues-unsubscr...@hadoop.apache.org For additional commands, e-mail: yarn-issues-h...@hadoop.apache.org
